Phylogenetic Studies of Pantherine Cats (Felidae) Based on Multiple Genes, with Novel Application of Nuclear -Wbrinogen Intron 7 to Carnivores

Molecular Phylogenetics and Evolution 35 (2005) 483–495 www.elsevier.com/locate/ympev

Phylogenetic studies of pantherine cats (Felidae) based on multiple genes, with novel application of nuclear -Wbrinogen intron 7 to carnivores

Li Yu a,b,c, Ya-ping Zhang a,b,¤

a Laboratory of Cellular and Molecular Evolution, and Molecular Biology of Domestic Animals, Kunming Institute of Zoology, Chinese Academy of Sciences, Kunming 650223, China b Laboratory for Conservation and Utilization of Bio-Resource, Yunnan University, Kunming 650091, China c Graduate School, Chinese Academy of Sciences, Beijing, China

Received 14 July 2004; revised 16 January 2005

Abstract

The pantherine lineage of the cat family Felidae (order: Carnivora) includes Wve big cats of genus Panthera and a great many midsized cats known worldwide. Presumably because of their recent and rapid radiation, the evolutionary relationship among pantherines remains ambiguous. We provide an independent assessment of the evolutionary history of pantherine lineage using two complete mitochondrial (mt) genes (ND2 and ND4) and the nuclear -Wbrinogen intron 7 gene, whose utility in carnivoran phylogeny was Wrst explored. The available four mt (ND5, cytb, 12S, and 16SrRNA) and two nuclear (IRBP and TTR) sequence loci were also combined to reconstruct phylogeny of 14 closely related cat species. Our analyses of combined mt data (six genes; t3750 bp) and combined mt and nuclear data (nine genes; t6500 bp) obtained identical tree topologies, which were well-resolved and strongly supported for almost all nodes. Monophyly of Panthera genus in pantherine lineage was conWrmed and interspeciWc aYnities within this genus revealed a novel branching pattern, with P. tigris diverging Wrst in Panthera genus, followed by P. onca, P. leo, and last two sister species P. pardus and P. uncia. In addition, close association of Neofelis nebulosa to Panthera, the phylogenetic redeWnition of Otocolobus manul within the domestic cat group, and the relatedness of Acinonyx jubatus and Puma concolor were all important Wnd- ings in the resulting phylogenies. The potential utilities of nine diVerent genes for phylogenetic resolution of closely related pantherine species were also evaluated, with special interest in that of the novel nuclear -Wbrinogen intron 7.  2005 Elsevier Inc. All rights reserved.

Keywords: -Fibrinogen intron; Pantherine lineage; Felidae; Phylogenetic analysis

1. Introduction The pantherine lineage, as the most recently evolved (within 1–8 MYA; Janczewski et al., 1995; Pecon Slat- The Felidae, or cat family, is characterized by recent tery et al., 1994) and largest felid group (around 20 cat bursts of diversiWcation within the last 10–15 million species; Janczewski et al., 1995) has demonstrated great years (Johnson and O’Brien, 1997; Martin, 1980; confusion in their taxonomy and phylogeny. These Nowak, 1999; Werdelin, 1985). Thirty-eight cat species pantherine cats consist of Wve big cats of genus Panthera of this family are generally divided into the pantherine, and a great many midsized cats species. They had been domestic cat, and ocelot lineages (Ewer, 1973; Janczew- disputably assigned to 2–13 genera under various classi- ski et al., 1995; Leyhausen, 1979; Masuda et al., 1996). Wcation schemes in past studies (Ewer, 1973; Hemmer, 1978; Leyhausen, 1979; Nowak, 1999) and moreover, ¤ Corresponding author. Fax: +86 871 5195430. phylogenetic relationships among these pantherine spe- E-mail address: [email protected] (Y.-p. Zhang). cies have also been controversial.

A wealth of molecular characters have been used to elis temminckii (Asiatic golden cat), Prionailurus bengal- decipher feline evolutionary history, including protein ensis (Asiatic leopard cat), Lynx lynx (lynx), Puma electrophoresis, allozyme data, karyology, endogenous concolor (puma), and Acinonyx jubatus (cheetah) were retroviruses, mitochondrial (mt) DNA sequences, sex included. In addition, two representatives of the domes- chromosomes-linked genes, and chemical signals (Bin- tic cat lineage and one outgroup taxa from family Viv- inda-Emonds, 2001; Collier and O’Brien, 1985; Johnson erridae or Hyaenidae were also utilized. Taxonomic and O’Brien, 1997; Lopez et al., 1994; O’Brien et al., classiWcation of cat species followed Nowak (1999). 1987; Pecon Slattery and O’Brien, 1998; Reeves and Total genomic DNA was isolated from blood, frozen or O’Brien, 1984). However, little prior research focused hair tissues based on the method of Sambrook et al. nuclear genes at the DNA level. In the present paper, the (1989) and prepared for subsequent polymerase chain seventh intron of the single-copy Wbrinogen gene reaction (PCR). (-chain; -Wbrinogen intron 7) from the nuclear genome Two felid-speciWc primers FGB-FelF and FGB-FelR was used for phylogenetic resolution among closely (Table 2) used to amplify -Wbrinogen intron 7 related pantherine cats. The utility of this gene segment (t650 bp) were designed from conserved regions in has been successfully explored at diVerent taxonomic Xanking exons, based on the homologous comparisons levels in studies of birds (Johnson and Clayton, 2000; of available -Wbrinogen sequences for birds (Prychitko Moyle, 2004; Prychitko and Moore, 1997, 2000, 2003; and Moore, 1997), human (Chung et al., 1983), and rat Weibel and Moore, 2002) and reptiles (Creer et al., 2003; (Eastman and Gilula, 1989). MtDNA PCR products Giannasi et al., 2001), however, still lacking in those of were obtained using ND2-FelF/ND2-FelR primer pair mammals. Our work is the Wrst to explore the potential for ND2 (1038 bp), as well as ND4-FelF/ND4-FelR of -Wbrinogen intron 7 as a genetic marker in carnivo- primer pair for ND4 (1368 bp). Additional internal ran systematics. We also sequenced two large complete primers for amplifying these three genes (Table 2) were NADH dehydrogenase (ND2 and ND4) genes from mt also synthesized. The optimal conditions adopted in genome, given the general thought that analysis of multi- PCRs were 95 °C initial hot start for 5 min, 35 cycles of ple independently inherited genes is especially eVective in 94 °C denaturation for 1 min, 50–56 °C annealing for testing for congruence and estimating organismal phy- 1 min, and 72 °C extension for 1 min. logeny and, in addition, our previously reported nuclear With above-mentioned primers, three new target seg- interphotoreceptor retinoid-binding protein (IRBP) and ments for each sample were acquired in almost all cases transthyretin (TTR) genes (Yu et al., 2004b), together except an incomplete ND2 sequence (683 bp) from P. with four other available mtDNA characters (12SrRNA, uncia (snow leopard) due to sequencing diYculty in one cytochrome b, 16SrRNA, and ND5 genes; Janczewski end of the gene, as well as unavailable -Wbrinogen et al., 1995; Johnson and O’Brien, 1997; Masuda et al., intron 7 sequences from P. onca (jaguar) and Puma con- 1996) for the same set of cat species were also added to color (puma) because of insuYcient template content in the present analyses. hair tissues as well as from Acinonyx jubatus (cheetah) Sequence data from 9 genes (three nuclear and six mt) because of the absence of sample. These three taxa were of 12 pantherine cats and 2 domestic cats was analyzed, thus only represented in the mtDNA datasets. ND2 and separately or in a variety of combinations here, with a ND4 sequences for Felis catus (domestic cat) and Acin- view of (1) providing a broader understanding of inter- onyx jubatus (cheetah) were directly retrieved from Gen- speciWc relationships within the pantherine group, (2) Bank. In all, 34 out of 40 ingroup sequences were assessing the utility of -Wbrinogen intron 7 as a novel produced in this study. marker in carnivoran phylogenetics, and (3) comparing evolutionary patterns of diVerent genes and their values 2.2. Sequencing and data analysis for resolution of low level feline questions. The ampliWed PCR products were puriWed and sequenced in both directions with an ABI automated 2. Materials and methods sequencer. Acquired sequences were submitted to Gen- Bank for BLAST searching (Altschul et al., 1997) to ver- 2.1. DNA samples and PCR ampliWcations ify the data. New GenBank Accession Nos. are given in Table 1. Fourteen felids in the family Felidae were examined Separate alignments of -Wbrinogen intron 7 (from 11 and listed in Table 1. All the Wve currently recognized ingroup) and two mtDNA (ND2 and ND4; from 14 members of genus Panthera in the pantherine lineage, ingroup) sequences data were conducted with CLUS- including P. leo (lion), P. tigris (tiger), P. pardus (leop- TAL X program (Thompson et al., 1997) and reWned by ard), P. onca (jaguar), and P. uncia (snow leopard), and visual inspection. Protein-coding mtDNA alignments seven other pantherine cats, including Neofelis nebulosa were straightforward while the -Wbrinogen intron 7 (clouded leopard), Otocolobus manul (Pallas’s cat), Prof- alignment exhibited obvious sequence length variation. L. Yu, Y.-p. Zhang / Molecular Phylogenetics and Evolution 35 (2005) 483–495 485 C NC_001700 AF006402 NC_001700 AF006401 ). AY525058 AF039724 AY525057 Z11811 AY170072 AY525033 NC_001700 AY634402 Janczewski et al., 1995; Masuda et al., 1996 ND2 ND4 exon IRBP intron TTR 16SrRNA ND5 NC_001700 AY170042 AY634390 -Fibrinogen AY634371 AY634383 AY634395 AY525041 AY525064AY634369 AF006443 AF006444 AY634381 AY634393 AY525034 AY525059 AF006447 AF006448 intron AY634379 AY634378 erent families (b) were utilizedfor the samegenes. V . China Province, Yunnan of South China Yunnan Province, China AY634376 AY634388 AY634400 AY525035 AY525060 AF006437 AF006438 GuangxiChina Province, QinhaiChina Province, Yunnan Province, China AY634380 AY634405 AY525040 AY525055 Nowak (1999) cation of of cation W LynxPumaCheetah Xinjiang Province, China Xian Zoo,China AY634377 AY634389 AY634401 AY525038 AY525062 AF006413 AY634392 AF006414 AY634404 AY463959 AY463959 AF006455 AF006456 AY463959 AY463959 PantherLionTigerJaguar Province, Yunnan of South Snow leopardClouded leopardPallas’s cat China Yunnan Province, China Kunming Zoo, ChinaAsiatic golden Yunnan Province, China Xian Zoo,China cat AY634373Asiatic leopard Xining Zoo, China AY634372 AY634374cat AY634385 AY634384 AY170043 AY634397 AY634396 AY634398 AY634375 AY525032 AY525037 AY634370 AY525036 AY525056 AY634387 AY525061 AF039725 AF006425 AY634382 AY634399 AF006459 AF006547 AY634391 AF006426 AY634394 AY525039 AF006460 AF006458 AY634403 AY525063 AY525042 AY525065 AF006431 AF006449 AF006432 AF00 6450 AF006441 AF006442 Common nameCommon source Sample gene Nuclear genes Mitochondrial gene Nuclear genes Mitochondrial cat Domestic Wild cat Chinese desert cat palm Masked civet Spotted hyena AY170057 AY170087 AF039728 AF006391 AF006392 erent speciessame of the genus(a) or di V b b A a a c name W Lynx lynx Lynx Puma concolor Acinonyx jubatus Panthera pardus Panthera leo tigris Panthera Panthera onca Panthera uncia Neofelis nebulosa Otocolobus manul Profelis temminckii Prionailurus bengalensis Crocuta crocuta TaxonScienti Newly established datasets Previouslyavailable datasets Felis catus Felis silvestris Felis bieti Paguma larvata B B Taxonomic denomination followed classi followed denomination Taxonomic di to belonging taxa cases, two In ( publications from obtained were sequences their and unavailable are genes cytb and 12SrRNA of numbers Accession cat lineage lineage B C A Domestic Domestic Pantherine Outgroups Table 1 Table study this in used sequences and samples taxonomic of List 486 L. Yu, Y.-p. Zhang / Molecular Phylogenetics and Evolution 35 (2005) 483–495

Table 2 List of PCR primers for ampliWcation and sequencing of nuclear -Wbrinogen intron, mt ND2, and ND4 genes for felids Target gene Primer namea Sequence (5Ј–3Ј) -Fibrinogen intron External primers FGB-FelF CACAACGGCATGTTCTTCAGCACG FGB-FelR TACCACCATCCACCACCATCTTCTT Internal primers FGB-FelF397 ATCACTCTAAGACGTTTCCACTAT FGB-FelR867 CCTCAGTGTTAGAGAACCCTGGAA mtND2 External primers ND2-FelF CCATACCCCGAAAATGTTGGTTTAT ND2-FelR AGCTTTGAAGGCTCTTGGTCT Internal primers ND2-FelF337 TTCTGAGTGCCCGAAGTCACACAAGG ND2-FelR689 GTTTTATTTCATGTTTGTGATA mtND4 External primers ND4-FelF GGCACTGACTATGTACAAAACCT ND4-FelR GACAGCATCTAGAACTTTGACCAT Internal primers ND4-FelF275 AAAAACTATACATCACAATACT ND4-FelR1066 CTGTAGCTCCTATATAGCTTCAGGG ND4-FelF362 TATTTGAAGCCACATTAATCCC ND4-FelR916 GCTAGTAGTCATCAGGCAGC a Numbers in the primer names refer to the position of the 5Ј end of the primer from alignment of sequences.

Besides one single-nucleotide deletion unique for Prof- ratio, the proportion of invariable sites, and rate hetero- elis temminckii and two small indels in the outgroup geneity among sites were considered. Bayesian analyses taxa, we found two unusual 239 bp long insertions, with started with randomly generated trees and four Markov 87.9% sequences similarity, emerged at diVerent posi- chains under default heating values were run for 2 £ 106 tions of Profelis temminckii and Otocolobus manul - generations, with sampling at intervals of 100 genera- Wbrinogen intron region, respectively (hereafter referred tions. To ensure that the analyses were not trapped on as Gol-ins and Pal-ins). Prior to phylogenetic analyses, local optima, the dataset was run three times indepen- Gol-ins and Pal-ins were removed from the original dently. We determined the burn-in period by checking alignment, leaving 657 nucleotide positions in the new for likelihood stationarity. To test for nodal reliabilities, nuclear alignment. heuristic bootstrap analyses (Felsenstein, 1985; 1000 rep- Pairwise distances based on the Tamura and Nei licates for MP and 100 for ML, respectively) and poster- (1993) (TN93) method were calculated with program ior probabilities were applied, with groups appearing in MEGA (Kumar et al., 2001). Stationary of nucleotide 50% or more of the trees in bootstrap analysis and sam- composition across taxa were examined using the 2 test pled with 80% or more frequency in Bayesian analysis in PAUP*4.0b8 (SwoVord, 2001). We also plotted the retained. number of transitions and transversions against TN93 distance to measure saturation patterns for all sites and 2.3. Combination with other available datasets each codon position using DAMBE (Xia, 2000). The assumption of molecular clock was tested using the Available sequences data from four other mt genes method of relative-rate test with program PHYLTEST 12SrRNA (352–354 bp), cytochrome b (cytb; 289 bp), (Kumar, 1996). 16SrRNA (372–375 bp), and ND5 (318 bp) (Janczewski Phylogenetic analyses were performed under maxi- et al., 1995; Johnson and O’Brien, 1997; Masuda et al., mum parsimony (MP), maximum likelihood (ML) and 1996) for the 15 taxa in the ND2 and ND4 datasets, as newly developed Bayesian (Larget and Simon, 1999) cri- well as those from two other nuclear genes IRBP exon teria. Analyses were conducted using PAUP*4.0b8 (1274 bp) and TTR intron (784–839 bp) (Yu et al., (SwoVord, 2001) and MrBayes 3.0B4 (Rqnquist and 2004b) for the same 12 taxa as those of present -Wbrino- Huelsenbeck, 2003). ND2 gene trees were rooted using gen intron 7 dataset, were also obtained from the Gen- Crocuta crocuta (spotted hyena; Yoder et al., 2003) as Bank database and combined in a variety of ways for the outgroup while -Wbrinogen intron 7 and ND4 trees analyses (9 in total). using Paguma larvata (masked palm civet) because no Phylogenetic reconstruction for six mt sequences (i.e., sequences of these two genes could be obtained from 12SrRNA + cytb + 16SrRNA + ND5 + ND2 + ND4), spotted hyena. In MP analysis, a branch-and-bound three nuclear sequences (i.e., -Wbrinogen +IRBP + TTR), search strategy was employed with all characters treated and combined sequences data were performed following as equal weights and gaps as missing data. The the aforementioned methods (MP, ML, and Bayesian best-Wtting models of sequences evolution for ML and algorithms). Outgroup sequences for Crocuta crocuta Bayesian analyses were estimated using program MOD- taxa were retrieved from GenBank in each data analysis, ELTEST 3.06 (Posada and Crandall, 1998), in which excepting those of -Wbrinogen and ND4 data as parameters of base frequencies, transition/transversion explained above. Partitioned Bremer support analysis L. Yu, Y.-p. Zhang / Molecular Phylogenetics and Evolution 35 (2005) 483–495 487

(PBS; Bremer, 1988, 1994), as implemented by Tre- illustration of species-speciWc SINE family members eRot.v2 (Sorenson, 1999), was used to measure the located in intronic regions and it appeared that the Gol- respective contribution of each gene to the total nodal ins and Pal-ins were derived from two independent inser- Bremer support. tion events. Evolutionary mechanisms underlying in Three of 14 ingroup taxa (P. onca, Puma concolor, these two repeated elements are unclear but deWnitely and Acinonyx jubatus) in mtDNA dataset were not rep- merit further investigation. resented by nuclear data, and as a result, their nuclear characters in the combined analyses using all evidence 3.2. Base composition and substitution patterns were coded as missing data. We also tested for incongru- ent phylogenetic signal between nuclear and mt genes Base composition was AT-rich biased in -Wbrinogen using the partition homogeneity test (Farris et al., 1994, (60.8%), ND2 (61.8%), and ND4 (60.1%) genes. No evi- 1995) in PAUP*4.0b8 (SwoVord, 2001). dence of base frequencies heterogeneity across taxa was observed for any genes. Nuclear -Wbrinogen gene showed the lowest transition/transversion ratio (ti/tvD1.94), con- 3. Results trasted with obvious tendency of transitions in the mtDNA genes (5.25 in ND2 and 8.41 in ND4). TN93 dis- 3.1. Novel Wndings of SINE element in -Wbrinogen tances within the ingroup taxa ranged from 0 to 4.6% for intronic segment -Wbrinogen (2.5% on average), from 1.4 to 18.4% for ND2 (12.9% on average), and from 0.6 to 19.3% for ND4 As mentioned above (see Section 2), two 239 bp large (15.3% on average). Comparisons of sequence divergences insertions (Gol-ins and Pal-ins) have occurred uniquely revealed faster rates of evolution in the ND2 and ND4 in Profelis temminckii (positions 96–334) and Otocolobus than in the -Wbrinogen gene at about 5.16 and 6.12 times, manul taxa (positions 587–825) -Wbrinogen intron 7, respectively. The results from DAMBE program (Xia, respectively. A BLAST search from GenBank database 2000) conWrmed a lack of multiple substitutions in each revealed that 225 bp of Gol-ins and Pal-ins exhibited the gene and codon position (plot not shown). Therefore, all highest similarities to carnivoran-speciWc short inter- substitutions in -Wbrinogen and those of the codon posi- spersed element, i.e., CAN SINE family. The results from tions in ND2 and ND4 genes were used in the analyses. the RepeatMasker program (A.F.A. Smit and P. Green, Sequence characteristics of individual genes are sum- unpublished data) conWrmed that they belong to the marized in Table 3. The six previously published datasets subfamily SINEC_Fc of CAN SINE. These two SINE- were included for comparison. The nine DNA sequences like insertions contained RNA polymerase III promoter including four protein-coding (ND2, ND4, ND5, and near the 5Ј end, a AT-rich region at the 3Ј end and per- cytb) and two rRNA genes (12S and 16S) from mt fect direct terminal repeats, characteristics of tRNA- genome, as well as two introns (-Wbrinogen and TTR) derived SINEs families (Coltman and Wright, 1994; and one exon (IRBP) from nuclear genome, range in size Okada, 1991a,b). Here, we unexpectedly Wnd another from 289 (cytb) to 1368 (ND4), with ND5 showing the

Table 3 Summary statistics for nuclear and mitochondrial gene segments used in this study Nuclear datasets Mitochondrial datasets -Fibrinogen IRBP TTR ND2 ND4 ND5 cytb 12SrRNA 16SrRNA intron exon intron Aligned sites (%) 657 1274 841 1038 1368 318 289 354 380 A% 28.4 18.2 28.6 36.7 32.3 33.5 27.8 36.5 33.6 C% 20.2 31.8 21.9 28.9 28.1 27.4 28.8 22.3 22.4 G% 19 32 19.5 9.3 11.8 7.8 15.5 17 20.1 T% 32.4 18 30 25.1 27.8 31.2 27.9 24.2 23.8 Variable sites 133 102 98 418 536 136 103 63 84 (20.24%) (8.01%) (11.65%) (40.27%) (39.18%) (42.77%) (35.64%) (17.8%) (22.11%) Parsimony- 28 21 13 251 385 80 66 26 37 informative sits (4.26%) (1.65%) (1.55%) (24.18%) (28.14%) (25.16%) (22.84%) (7.34%) (9.74%) ti:tv ratio 1.94 3.93 1.74 5.25 8.41 5.48 7.09 3.28 3.98 Mean TN distance (%) 2.5 (0–4.6%) 1.1 1.1 12.9 15.3 14.8 12.3 3.9 5.6 within ingroup (0.2–2.3) (0–1.9) (1.4–18.4) (0.6–19.3) (0–23.2) (1.8–17.5) (0.6–6.1) (1.4–7.9) Proportion of 0 0.8081 0.5457 0.4931 0.5612 0 0 0 0 invariable sites (I) Gamma-shape 0.401 11 1.419 1.8889 0.3913 0.1808 0.1615 0.1752 parameter () 488 L. Yu, Y.-p. Zhang / Molecular Phylogenetics and Evolution 35 (2005) 483–495 highest percentage of variable sites (42.77%) and IRBP other nodes, however, received stronger supports from the lowest (8.01%). ND4 had the most parsimony-infor- the combined data. The 11 cat species were separated mative sites (28.14%) while TTR had the least (1.55%). into two large clades: one containing all representatives Pairwise divergences comparisons revealed that among of Panthera and its closest sister species Neofelis nebul- nuclear genes the substitution rate of -Wbrinogen was osa while the other contained the remaining pantherine about two times higher than those of both IRBP and cats and the two species of domestic cat lineage. MP TTR. Among mt genes, ND4 showed the fastest rate of bootstrap and Bayesian analyses recovered similar, but substitution, followed by ND5, ND2, cytb, 16S, and 12S. less-resolved trees compared to the ML analysis (- As expected, estimates of ti/tv ratio were higher for mt Wbrinogen MP tree length D 157, CI D 0.879, and than nuclear DNA genes. In addition, mt sequences were RC D 0.626; combined MP tree lengthD 380, CI D 0.903, generally much more divergent than nuclear sequences and RC D 0.650) and notably diVered in placing P. tem- (8.2 times in substitution rate on average), and they also minckii as the earliest diverging cat species in both MP showed greater rate heterogeneity among sites (lower trees (52–53%). O. manul and L. lynx were paired estimates of ) (Table 3). Among the nine examined together in all analyses of -Wbrinogen dataset (MP, genes, the relative rate estimation of cytb over -Wbrino- 61%; ML, 58%; Bayesian, 97%). Combined nuclear data gen gene has often been measured in previous avian analyses consistently identiWed closest aYnity between studies. For the feline species examined in this study the P. uncia and P. pardus (MP, 93%; ML, 81%; Bayesian, value was 4.92, compared with 2.79 and 3.7 in various 99%) within Panthera, with P. leo placed as the sister woodpeckers (Prychitko and Moore, 1997, 2000) and 5.6 taxon to them (MP, 96%; ML, 83%; Bayesian, 100%). among pigeons and doves (Johnson and Clayton, 2000). The genus Panthera and N. nebulosa formed a monophyletic group in all nuclear analyses. 3.3. Phylogenetic analyses based on nuclear characters 3.4. Phylogenetic analyses based on mitochondrial ML bootstrap trees based on analyses of individual - characters Wbrinogen dataset and combined nuclear dataset (- Wbrinogen + IRBP + TTR; 2772 bp) are shown in Figs. ML bootstrap trees derived from analyses of individ- 1A and B, respectively. The topologies of two gene trees ual ND2 and ND4 datasets are shown in Figs. 2A and B, are identical in all respects except that the weakly sup- respectively. The topologies of the two gene trees con- ported sister-group relationship between Otocolobus trasted in several aspects; however, in no case was a node manul and Lynx lynx (58%) inferred in the -Wbrinogen with a strong bootstrap value for one gene contradicted tree was not supported in the combined nuclear tree. All by a well-supported node for the other gene. The ND4

Fig. 1. Bootstrap maximum likelihood (ML) trees for -Wbrinogen intron gene (A; ¡ln l D 1679.5955) and combined -Wbrinogen, IRBP and TTR genes (B; ¡ln l D 5857.64 61) from 11 species of felids, using Paguma larvata as outgroup. The numbers above branches indicate bootstrap values, of which only those greater than 50% are shown. The maximum parsimony (MP) and Bayesian analyses obtained similar tree topologies. L. Yu, Y.-p. Zhang / Molecular Phylogenetics and Evolution 35 (2005) 483–495 489

Fig. 2. Bootstrap maximum likelihood (ML) trees for mt ND2 (A; ¡ln l D 5032.1025) and ND4 genes (B; ¡ln l D 6799.0806) for 14 species of felids, using Crocuta crocuta and Paguma larvata as outgroups, respectively. The numbers above branches indicate bootstrap values, of which only those greater than 50% are shown. The maximum parsimony (MP) and Bayesian analyses obtained similar tree topologies. tree tended to have overall greater resolution and higher with mt sequences from this study and analyzed simul- nodal supports but lower RC indices. In ND4 analyses, taneously (t3750 bp). The topologies obtained by vari- complete resolution among Wve Panthera members and ous analytic methods were identical (MP tree their close relationship with N. nebulosa were supported. length D 2808, CI D 0.531, and RC D 0.228) and exhib- P. uncia paired with P. pardus (100%) rather than with ited much improved resolution and nodal support than P. onca (60%) as identiWed in ND2 data. The precise either mtDNA gene region alone. ML analysis of the placements of the other cats were ambiguous in the two combined mt data is presented in Fig. 3A. This multiple gene trees as demonstrated by the presence of polyto- gene tree is completely resolved and strongly supported mies and low bootstrap values. MP (tree length D 882, for all nodes except for the position of P. temminckii CI D 0.569, and RC D 0.251) and Bayesian trees from (53%). Similar support values for this node were also ND2 data analyses provided a distinct picture from Fig. obtained by MP (50%) and Bayesian (75%) analyses. 2A, with N. nebulosa either being the most basal taxon of The 14 cat species split into two large groups in all anal- all the felids in MP tree (80%) or an uncertain taxon in a yses, one containing the monophyletic Panthera genus polytomy at the base of Bayesian tree. In addition, O. and N. nebulosa (MP, 56%; ML, 81%; Bayesian, 84 %) manul was weakly placed as the sister taxon to the two and the other containing the remaining pantherine cats species of domestic cat lineage in MP tree (52%). P. tem- and two domestic cat lineage species (MP, 94%; ML, minckii, P. concolor, and A. jubatus were grouped 98%; Bayesian, 100%). In the former group, N. nebulosa together in the Bayesian tree although with low poster- occupied the most basal position followed by P. tigris, ior probability (82%). In the ND4 gene analyses, MP and P. onca, P. leo, and last two most recently diverged sis- Bayesian trees were largely congruent to Fig. 2B (MP ter species P. pardus and P. uncia. In the latter, all felids tree length D 1305, CI D 0.490, and RC D 0.235). Internal were further subdivided into two clades: one contained relationships within Panthera and its sister-group of N. P. temminckii, P. concolor, A. jubatus plus L. lynx (MP, nebulosa remained well-supported while other cat species 66%; ML, 93%; Bayesian, 99%) and the other contained either constituted a monophyletic clade under the Bayes- O. manul, P. bengalensis plus the two domestic cat spe- ian method (91%), as was also seen in Fig. 2B, or col- cies (MP, 50%; ML, 79%; Bayesian, 100%). P. concolor lapsed into a large polytomy under MP method. and A. jubatus were strongly paired together, with P. Four mt genes (12SrRNA, cytb, 16SrRNA, and temminckii and L. lynx successively joining the Wrst ND5) for the same set of cat species were combined subclade. In the other subclade O. manul consistently 490 L. Yu, Y.-p. Zhang / Molecular Phylogenetics and Evolution 35 (2005) 483–495

Fig. 3. Bootstrap maximum likelihood (ML) trees for combined mt (A; ¡ln l D 16744.44364) and combined mt and nuclear genes (B; ¡ln l D 22187.8346) from 14 species of felids, using Crocuta crocuta as outgroup. The numbers above branches indicate bootstrap values, of which only those greater than 50% are shown. The maximum parsimony (MP) and Bayesian analyses obtained nearly identical tree topologies. showed a robust association with two domestic cat rep- divergence. The majority of variations and informative resentatives and P. bengalensis was identiWed as the sis- sites in the total dataset are from mt regions, especially ter taxon to them. protein-coding genes. Nuclear genes only account for 20% of the variable and 7% of the informative charac- 3.5. Phylogenetic analyses based on entire characters ters, though they exhibit lower levels of homoplasy. It thus seemed that mt genes are superior to nuclear genes The partition homogeneity test (Farris et al., 1994, for tracing phylogenetic relationships among closely 1995) indicated the homogeneity of phylogenetic signal related felines. In our view, the hypotheses based on between our nuclear and mt datasets (p D 0.98) allowing combined mt or full datasets (Fig. 3) represent the best a total of about 6500 bp characters to be combined for current estimate of feline phylogeny in the present phy- tree reconstruction. All analyses resulted in essentially logenetic reconstruction because all nodes are resolved identical topologies as those based on combined mt data and highly supported. alone with similar or slightly improved support for most nodes and higher RC index (0.243 vs. 0.228). The ML bootstrap tree from this dataset is presented in Fig. 3B. 4. Discussion Broken lines in the tree denote the branches connecting three taxa P. concolor, A. jubatus, and P. onca, from 4.1. Monophyly and interspeciWc relationships of panthera whom nuclear sequences were missing. genus Partitioned Bremer values (Table 4) reveals that the majority of phylogenetic information from the com- Traditionally, Panthera is composed of Wve recog- bined gene-based topologies was contributed by the mt nized extant species. However, interspeciWc relationships characters, with ND4 gene holding the highest percent- within the genus have been equivocal probably due to age (43.83%), followed by 16SrRNA (16.05%), ND2 and extremely recent speciation (1–2 MYA; Kurten and cytb equally (11.11%), 12SrRNA (8.64%), and ND5 Anderson, 1980; Wayne et al., 1989) resulting in short (8.02%), in sharp contrast to the nuclear characters internodes that have not been resolved because of (1.23% in total). The relatively weak inXuence of nuclear insuYcient sequence data used in previous studies (see partition upon analyses of complete set of genes was Fig. 4). In this paper, although individual genes exam- unsurprising, given their extremely low degree of genetic ined failed to provide a well-supported phylogeny, the L. Yu, Y.-p. Zhang / Molecular Phylogenetics and Evolution 35 (2005) 483–495 491

analyses of full data and combined mt data (Fig. 3), con-

B) sistently produced a robust, well-resolved tree. Our phylogenies strongly uphold the monophyly of Panthera. Fig. 4 The internal aYnities within the genus are shown to be novel, and our results diVer from all previous hypotheses (Fig. 4). Though one of the species, P. onca, was not included in present nuclear datasets, the branching Combined nuclearand mt gene trees( orders of the other four (Fig. 1) were mostly congruent between mt, nuclear and combined analyses, the excep- tion being in the poorly supported result obtained from

112 the ND2 gene (Fig. 2A). In conclusion, our analyses ¡ Combined nuclear clearly indicate that P. tigris is the sister taxon to the other members of Panthera, within the clade containing the other members of the genus, P. unica and P. pardus are sister species, next joined by P. leo and Wnally by P. onca. The most interesting and novel Wnding of this study is the strong sister-group pairing of P. uncia and P. pardus. b

et) Morphological studies and some previous molecular work favored placing P. uncia as the most basal taxon in the genus (Figs. 4B–D). A recent analysis of chemical signal, however, suggested it is the closest relative of P. 1 2 (1.23%) 1 (0.62%) 2 (1.23%) 162 100 -Fibrinogen IRBP TTR tigris (Fig. 4E). P. pardus has also been alternatively ¡ ¡ Nuclear datasets hypotheized as the sister taxon of P. leo (Figs. 4A, C, E, and F) or of P. onca from protein electrophoresis data (O’Brien et al., 1987), or based on combined morphology, karyology and molecular sequences as a diverging taxon after P. uncia (Fig. 4D). Our analyses also ques- tion fossil evidence that suggests a close relationship of P. leo with P. tigris or P. onca (Hemmer, 1971; NeV, ) were unavailable, the PBS scores for the nuclear datasets would be underestimated in some aspects. some in underestimated be would datasets nuclear the for scores PBS the unavailable, were ) 1982). W 414 1200002 Taking all evidence together, we can nd that no two ¡ ¡ analyses have come to a completely identical result regarding aYnities among these Wve species, with conclu- sions varying depending on the analytic methods and Acinonyx jubatus character type used. However, given the remarkable con- 43928010129 21210000010 69 900009 92 000000 11 300003 65 100001 , and , and ¡ ¡ ¡ ¡ ¡ ¡ gruence between genes of a distinct linkage group and the large size of our dataset, as well as the robust support of many nodes, we consider our gene tree the preferred

4 4 1 00 0 12 28 3 120 interpretation of Panthera species relationships. Both ¡ ¡ ¡ ¡

Puma concolor Wilcoxon signed-ranks (Templeton, 1983) and Shimoda- , ira–Hasegawa tests (as implemented in PAUP*) suggest that the combined mt or full data topology we recovered P. onca (Fig. 3) were signiWcantly the best estimate for present dataset. None of six prior hypotheses (Fig. 4) was sup-

B. ported (p < 0.001, data not shown).

Fig. 3 4.2. Phylogenetic resolutions of non-panthera species in 30 13

¡ ¡ pantherine lineage

As with Panthera, dispute also exists over the taxonomy and phylogeny of the non-Panthera panthe- 2101 23000023 225675 ¡ ¡ Mitochondrial datasets cytb 12SrRNA 16SrRNA ND5 ND2 ND4 mt Combined rines, most recently reviewed by Mattern and McLennan (2000) and Bininda-Emonds (2001). Seven a

Nodes are numberedas in Nuclear sequence dataof three samples ( species, each representing one genus of non-panthera a b 51 2 2 2 2 354 1 4 6070 3 080 1 2 9 2 0 1 0 13 0 0 3 2 0 1 1 2 9 5 1 30 0 55 0 0 0 0 55 10 Total 18 (11.11%) 14 (8.64%) 26 (16.05%)(8.02) 13 18 (11.11%) 71 (43.83%) 160 (98.77%) 10 1 3 0 13 1 1112 1 1 1 0 1 0 Table 4 Table datas mt and nuclear (combined tree MP evidence total the on node each for analyses (PBS) support Bremer partitioned of Results Nodes pantherines (Ewer, 1973; Nowak, 1999), were sampled 492 L. Yu, Y.-p. Zhang / Molecular Phylogenetics and Evolution 35 (2005) 483–495

Fig. 4. Competing hypotheses of phylogenetic relationships within Panthera deduced from (A) partial 12SrRNA and cytb genes (Janczewski et al., 1995), (B) partial 16SrRNA and ND5 genes (Johnson and O’Brien, 1997), (C) supertree construction of data sources spanning 25 years of systematic research (Bininda-Emonds et al., 1999), (D) partial 12SrRNA, 16SrRNA, ND5, and cytb genes with morphological and karyological characters (Mattern and McLennan, 2000), (E) chemical signals (Bininda-Emonds, 2001), and (F) complete mtDNA control region (Jae-Heup et al., 2001). in this study. Although precise relationships among 4.2.2. Phylogenetic position of Otocolobus manul them were not well resolved in separate gene analyses, Within non-Panthera pantherines, one cat species that the combined mt and full data, however, provided long has been considered diYcult to place in phyloge- insightful understanding of these pantherines’ evolu- netic studies is Otocolobus manul (Pallas’s cat). Studies of tion. morphology (Wayne et al., 1989), albumin immunological distance (AID; Collier and O’Brien, 1985), karyology 4.2.1. Close association of Neofelis nebulosa to Panthera (Wurster-Hill and Centerwall, 1982), mt cytb (Masuda The clustering of N. nebulosa with Panthera was well et al., 1996), and ND5 genes (Johnson and O’Brien, corroborated by the present data. It also supported the 1997) identiWed the lineage resulting in O. manul as an increasing evidence that N. nebulosa was included as the early divergence among the domestic cat lineages. How- most basal member within Panthera genus (Bininda- ever, sequences from mt 12S and 16SrRNA genes (John- Emonds et al., 1999; Johnson and O’Brien, 1997; Mat- son and O’Brien, 1997; Masuda et al., 1996), together tern and McLennan, 2000; Yu et al., 2004b). The possi- with nuclear characters (Pecon Slattery and O’Brien, ble sister species status of N. nebulosa and P. tigris, as 1998; Yu et al., 2004b), provided novel evidence for the proposed by Janczewski et al. (1995) from mt 12SrRNA inclusion of it in the pantherine lineage. and cytb, is not supported by our results. Such a group- Our individual solutions were equivocal on this issue, ing requires an extra 40 steps and signiWcantly worse in which the position of O. manul was poorly resolved (p < 0.0001) for the combined mt and full datasets. MP and showed no particular aYnities to any taxon except analyses of the ND2 gene alone, however, revealed that for those in -Wbrinogen analyses and one of ND2 anal- N. nebulosa was less closely linked with Panthera and ysis. All -Wbrinogen trees diagnosed the association of occupied a basal position in the tree (bootstrap O. manul with L. lynx with low bootstrap values (61% in value D 82%). This ideosyncratic pattern of relationship MP; 58% in ML) but high posterior probability (97%), can be explained by the “long branch attraction” eVect indicative of probable inclusion of it within the panther- between N. nebulosa and the outgroup taxon. PHYL- ine lineage. Of note, Pecon Slattery and O’Brien (1998) TEST (Kumar, 1996) results have observed a signiW- obtained the same relationship from introns of two cantly accelerated rate of evolution in N. nebulosa Y-linked genes. A recent analysis by us, also based on relative to the other felids (p < 0.05; data not shown). It nuclear genes (i.e., IRBP and TTR), failed to recover the has been shown that among tree-building methods, ML linkage between O. manul and L. lynx; however, we is superior to MP in producing reliable trees eVectively argued for a closer relationship of it with the pantherine in cases of long-branch attraction, and it is likely to be so cats (Yu et al., 2004b). The combination of three nuclear in this case. A reversion of N. nebulosa to the traditional genes (i.e., -Wbrinogen, IRBP, and TTR), unexpectedly placement in the ND2 MP tree only requires seven addi- showed no support of any view in this respect (Fig. 2B). tional steps and no statistically signiWcant diVerence In contrast, O. manul was supported as the closest sister between these alternative topologies was found taxon to two species of domestic cats lineage in ND2 MP (p D 0.1083). (52%) and both combined mt and full datasets analyses L. Yu, Y.-p. Zhang / Molecular Phylogenetics and Evolution 35 (2005) 483–495 493 with strong bootstrap supports (96 and 92% in MP; 84 Both genes evolved at a rapid rate and no sign of signiW- and 76% in ML) and high posterior probabilities (99 and cant saturation were observed. Separate analyses of the 90%). Our analyses support placing the elusive O. manul ND2 and ND4 genes produced congruent tree topolo- inside the domestic cat lineage, although this was not gies to those obtained from the combined mt and full supported by nuclear data. data. The ND4 gene was found to contribute the most phylogenetic signal while ND2, equivalent to cytb, only 4.2.3. Relatedness of Acinonyx jubatus and Puma lesser than ND4 and 16SrRNA to the Wnal tree (Fig. 4; concolor see Table 4). ND4 gene alone appeared enough to The result of sister grouping between two distinctive resolve Panthera relationships (including N. nebulosa) in monotypic genera, A. jubatus and P. concolor in our terms of complete resolution and strong nodal supports combined mt dataset was another important point in (Fig. 2B), whereas it did not serve well in the rest of the obtained phylogenies, although sequences from both pantherines. The other Wve mt genes included in this taxa were missing in the nuclear data. The exact place- study (ND2, ND5, cytb, 12S, and 16SrRNA) were more ments of these two species have been problematic. Bin- helpful in the recovery of relatively basal divergences inda-Emonds (2001), for example, has recommended and aYnities among genera, thus making the combina- particular attention of them in future felid systematics. tion of all mt genes ideal for providing persuasive evi- Several previous studies of DNA sequences (Johnson dence of the resolution to all parts of the trees. Among and O’Brien, 1997; Mattern and McLennan, 2000; Pecon the six mt genes examined, ND4 was shown the most Slattery and O’Brien, 1998) deWned a monophyletic appropriate while ND5 the least for the tree construc- Puma group within the cat family composed of A. juba- tion of closely related species. ND2, cytb, and the two tus, P. concolor, and H. yagouaroundi, the former two of rRNA genes were best able to resolve nodes in the mid- which were sampled here. dle range of divergence. Our result conWrmed a closer grouping of Puma and Nuclear genes have been shown to have a lower sub- Acionyx to each other than either was to the other felids, stitution rate and are less subject homoplasy than the mt a relationship also has been suggested by mt cytb and genes. Our study also provides valuable information on 12SrRNA gene (Janczewski et al., 1995) and by fossil evi- the utility of three nuclear genes, -Wbrinogen, IRBP, dence (Adams, 1979). ConXicting views which place A. and TTR, in the phylogenetic resolution of these panth- jubatus as either the sister taxon to the Panthera group erine cats. IRBP exon and TTR intron have been widely (Salles, 1992), or as the earliest genus to diverge within used for resolving relationships at a variety of taxo- pantherine lineage (Collier and O’Brien, 1985) were not nomic levels, particularly in carnivoran species (Flynn upheld by any of our analyses, nor was the hypothesis and Nedbal, 1998; Flynn et al., 2000; Sato et al., 2003, supporting P. concolor as either a close relative to the 2004; Yoder et al., 2003; Yu et al., 2004a,b). The -Wbrin- Panthera group (Salles, 1992) or as an divergent lineage ogen intron gene was newly explored in the present within modern felids (Bininda-Emonds, 2001). study to reconstruct phylogeny of felids, encouraged by the fact that this region has been shown to contain valu- 4.3. Utility of diVerent mt and nuclear genes in phylogeny able phylogenetic signals in various groups of avian taxa of cat family that diverged as anciently as 90 MYA as recently as 2–5 MYA (Prychitko and Moore, 1997, 2003). Although a Our results demonstrate that individual genes, includ- basal split dividing the sampled taxa into two large clus- ing the six previously published gene data (tree not ters and some aYnities within the clusters were revealed shown), and the combined nuclear dataset, fail to by ML analysis of the -Wbrinogen intron gene region recover a satisfying phylogeny. This lack of resolution is (Fig. 1A), the -Wbrinogen region displayed low largely due to insuYcient phylogenetic information in sequence divergence resulting in extremely short inter- individual loci. The concatenation of all mt genes, com- nodes, and consequently had less resolving power in prising four protein-coding and two rRNA genes, how- recovering relatively recent branching patterns in pan- ever, provides decisive resolution of the relationships therines. It is expected that this novel nuclear marker among the taxa examined, as does the dataset of com- would be better suited for resolving supergeneric (e.g., bined mtDNA and three nuclear genes. It is apparent interfamilial) relationships among carnivores and other that having more sites results in better resolution, and mammals. strong historical signals appear when multiple genes with distinct substitution dynamics are combined, as a consequence of their diVerential resolving powers at 5. Supplementary materials multiple taxonomic levels. Newly established ND4 and ND2 datasets in present The sequences reported in this paper have been study proved especially informative in phylogenetic deposited in the GenBank database. Accession Nos. reconstruction of these recently evolved pantherine cats. AY634369–AY634385 and AY634387–AY634405. 494 L. Yu, Y.-p. Zhang / Molecular Phylogenetics and Evolution 35 (2005) 483–495

Acknowledgments Jae-Heup, K., Eizirik, E., O’Brien, S.J., Johnson, W.E., 2001. Structure and patterns of sequence variation in the mitochondrial DNA con- This work was supported by grants from National trol region of the great cats. Mitochondrion 1, 279–292. Janczewski, D.N., Modi, W.S., Stephens, J.C., O’Brien, S.J., 1995. Natural Science Foundation of China (NSFC), Chinese Molecular evolution of mitochondrial 12S RNA and cytochrome b Academy of Sciences and Program for S & T Fund Pro- sequences in the pantherine lineage of Felidae. Mol. Biol. Evol. 12, ject of P.R. China (2001DEA10009-08). We thank Asso- 690–707. ciate Editor Dr. Derek E Wildman’s help for improving Johnson, W.E., O’Brien, S.J., 1997. Phylogenetic reconstruction of the the English of this manuscript. Felidae using 16SrRNA and NADH-5 mitochondrial genes. J. Mol. Evol. 44, S98–S116. Johnson, K.P., Clayton, D.H., 2000. Nuclear and mitochondrial genes contain similar phylogenetic signal for pigeons and doves (Aves: References Columbiformes). Mol. Phylogenet. Evol. 14, 141–151. Kumar, S., 1996. Phyltest: A Program For Testing Phylogenetic Adams, D.B., 1979. The cheetah: native American. Science 205, 1155– Hypothesis. Version 2.0. The Pennsylvania Statee University, Uni- 1158. versity Park. Altschul, S.F., Madden, T.L., SchäVer, A.A., Zhang, J., Zhang, Z., Kumar, S., Tamura, K., Jakobsen, I.B., Nei. M., 2001. MEGA2: Molec- Miller, W., Lipman, D.J., 1997. Gapped BLAST and PSI-BLAST: a ular Evolutionary Genetics Analysis Software. Version 2.1, Arizona new generation of protein database search programs. Nucleic Acids State University, Tempe, Arizona, USA. Res. 25, 3389–3402. Kurten, B., Anderson, E., 1980. Pleistocene Mammals of North Amer- Bininda-Emonds, O.R.P., Gittleman, J.L., Purvis, A., 1999. Building ica. Columbia University Press, NY. large trees by combining phylogenetic information: A complete Larget, B., Simon, D.L., 1999. Markov chain Monte Carlo algorithms phylogeny of the extant Carnivora (Mammalia). Biol. Rev. 74, 143– for the Bayesian analysis of phylogenetic trees. Mol. Biol. Evol. 16, 175. 750–759. Bininda-Emonds, O.R.P., 2001. The utility of chemical signals as phy- Leyhausen, P., 1979. Cat Behavior. Garland Press, NY. logenetic characters: an example from the felidae. Biol. J. Linnean Lopez, J.V., Yuhki, N., Masuda, R., Modi, W., O’Brien, S.J., 1994. Soc. 72, 1–15. Numt, a recent transfer and tandem ampliWcation of mitochondrial Bremer, K., 1988. The limits of amino acid sequence data in angio- DNA to the nuclear genome of the domestic cat. J. Mol. Evol. 39, sperm phylogenetic reconstruction. Evolution 42, 795–803. 174–190. Bremer, K., 1994. Branch support and tree stability. Cladistics 10, 295–304. Martin, L.D., 1980. Functional morphology and the evolution of cats. Chung, D.W., Que, B.G., Rixon, M.W.A., Mace Jr., M., Davie, E.W., Trans. Neb. Acad. Sci. 13, 141–154. 1983. Characterization of complementary deoxyribonucleic acid Mattern, M.Y., McLennan, D.A., 2000. Phylogeny and speciation of and genomic deoxyribonucleic acid for the chain of human Wbrin- Felids. Cladistics 16, 232–253. ogen. Biochemistry 22, 3244–3250. Masuda, R., Lopez, J.V., Pecon Slattery, J., Yuhki, N., O’Brien, S.J., Collier, G.E., O’Brien, S.J., 1985. A molecular phylogeny of the Felidae: 1996. Molecular phylogeny of mitochondrial cytochrome b and 12S immunological distance. Evolution 39, 437–487. rRNA sequences in the Felidae: ocelot and domestic cat lineages. Coltman, D.W., Wright, J.M., 1994. Can SINEs: a family of tRNA- Mol. Phylogenet. Evol. 3, 351–365. derived retroposons speciWc to the superfamily Canoidea. Nucleic Moyle, R.G., 2004. Phylogenetics of barbets (Aves: Piciformes) based Acids Res. 22, 2726–2730. on nuclear and mitochondrial DNA sequence data. Mol. Phyloge- Creer, S., Malhotra, A., Thorpe, R.S., 2003. Assessing the phylogenetic net. Evol. 30, 187–200. utility of four mitochondrial genes and a nuclear intron in the NeV, N.A., 1982. The Big Cats: The Paintings of Guy Coheleach. Asian pit viper genus, Trimeresurus: separate, simultaneous, and Abrams Press, NY. conditional data combination analyses. Mol. Biol. Evol. 20, 1240– Nowak, R.M., 1999. In: Walker’s Mammals of the World, sixth ed., vol. 1251. 2. Johns Hopkins University Press, Baltimore, MD. Eastman, E.M., Gilula, N.B., 1989. Cloning and characterization of a O’Brien, S.J., Collier, G.E., Benveniste, R.E., Nash, W.G., Newman, cDNA for the chain of rat Wbrinogen: evolutionary conservation A.K., Simonson, J.M., Eichelberger, M.A., Seal, U.S., Janssen, D., of translated and 3Ј-untranslated sequences. Gene 79, 151–158. Bush, M., Wildt, D.E., 1987. Setting the molecular clock in Felidae: Ewer, R.F., 1973. The Carnivores. Cornell Univ. Press, Ithaca, NY. the great cats, Panthera. In: Tilson, R.L., Seal, U.S. (Eds.), Tigers of Farris, J.S., Kallersjo, M., Kluge, A.G., Bult, C., 1994. Testing signiW- the World. Noyes, NJ, pp. 10–27. cance of congruence. Cladistics 10, 315–320. Okada, N., 1991a. SINEs: short interspersed repeated elements of the Farris, J.S., Kallersjo, M., Kluge, A.G., Bult, C., 1995. Constructing a eukaryotic genome. Trends Ecol. Evol. 6, 358–361. signiWcance test for incongruence. Syst. Biol. 44, 570–572. Okada, N., 1991b. SINEs. Curr. Opin. Genet. Dev. 1, 498–504. Felsenstein, J., 1985. ConWdence limits on phylogenies: an approach Pecon Slattery, J., Johnson, W.E., Goldman, D., O’Brien, S.J., 1994. using the bootstrap. Evolution 39, 783–791. Phylogenetic reconstruction of South American felids deWned by Flynn, J.J., Nedbal, M.A., 1998. Phylogeny of the Carnivora (Mamma- protein electrophoresis. J. Mol. Evol. 39, 296–305. lia): congruence vs incompatibility among multiple data sets. Mol. Pecon Slattery, J., O’Brien, S.J., 1998. Patterns of Y and X chromosome Phylogenet. Evol. 9, 414–426. DNA sequence divergence during the felidae radiation. Genetics Flynn, J.J., Nedbal, M.A., Dragoo, J.W., Honeycutt, R.L., 2000. 148, 1245–1255. Whence the red panda?. Mol. Phylogenet. Evol. 17, 190–199. Posada, D., Crandall, K.A., 1998. Modeltest: testing the model of DNA Giannasi, N., Malhotra, A., Thorpe, R.S., 2001. Nuclear and mtDNA substitution. Bioinformatics 14, 817–818. phylogenies of the Trimeresurus complex: implications for the gene Prychitko, T.M., Moore, W.S., 1997. The utility of DNA sequences versus species tree debate. Mol. Phylogenet. Evol. 19, 57–66. of an intron from the -Wbrinogen gene in phylogenetic analysis Hemmer, H., 1971. In: Zur Dharakerisierung und stratigraphischen bedut- of woodpeckers (Aves: Picidae). Mol. Phylogenet. Evol. 8, 193– ing von Panthera gombaszoegensis (Kretzoi, 1938). Neues Jahrbuch fur 204. Geologie und palaontologie, Monatchefte, pp. 701–711. Prychitko, T.M., Moore, W.S., 2000. Comparative evolution of the Hemmer, H., 1978. The evolutionary systematics of living Felidae: mitochondrial cytochrome b gene and nuclear -Wbrinogen intron 7 present status and current problems. Carnivore 1, 71–79. in woodpeckers. Mol. Biol. Evol. 17, 1101–1111. L. Yu, Y.-p. Zhang / Molecular Phylogenetics and Evolution 35 (2005) 483–495 495

Prychitko, T.M., Moore, W.S., 2003. Alignment and phylogenetic Templeton, A.R., 1983. Phylogenetic inference from restriction endo- analysis of -Wbrinogen intron 7 sequences among avian orders nuclease cleavage site maps with particular reference to the evolu- reveal conserved regions within the intron. Mol. Biol. Evol. 20, tion of humans and the apes. Evolution 37, 221–244. 762–771. Thompson, J.D., Gibson, T.J., Plewniak, F., Jeanmougin, F., Higgins, Reeves, R.H., O’Brien, S.J., 1984. Molecular genetic characterization of D.G., 1997. The clustalx windows interface: Xexible strategies for the RH-114 gene family of endogenous feline retroviral sequences multiple sequence alignment aided by quality analysis tools. of the domestic cat. J. Virol. 52, 164–171. Nucleic Acids Res. 24, 4876–4882. Rpnquist, F.R., Huelsenbeck, J.P., 2003. MrBayes 3: Bayesian phylo- Wayne, R.K., Benveniste, R.E., Janczewski, D.N., O’Brien, S.J., 1989. genetic inference under mixed models. Bioinformatics 19, 1752– Molecular and biochemical evolution of the carnivora. In: Gittle- 1754. man, J.L. (Ed.), Carnivore Behavior, Ecology, and Evolution. Cor- Salles, L.O., 1992. Felid phylogenetics: Extant taxa and skull morphol- nell Univ. Press, Ithaca, NY, pp. 465–495. ogy (Felidae, Aeluroidae). Am. Mus. Novit. 3047, 1–67. Weibel, A.C., Moore, W.S., 2002. A test of a mitochondrial gene-based Sambrook, E., Fritsch, F., Maniatis, T., 1989. Molecular Clonging. phylogeny of woodpeckers (Genus Picoides) using an independent Cold Spring Harbor Press, Cold Spring Harbor, NY. nuclear gene. Mol. Phylogenet. Evol. 22, 247–257. Sato, J.J., Hosoda, T., Wolsan, M., Tsuchiya, K., Yamamoto, M., Werdelin, L., 1985. Small Pleistocene felines of North America. J. Vert. Suzuki, H., 2003. Phylogenetic relationships and divergence times Paleo. 5, 194–210. among Mustelids (Mammalia: Carnivora) based on nucleotide Wurster-Hill, D.H., Centerwall, W.R., 1982. The interrelationships of sequences of the nuclear interphotoreceptor retinoid binding pro- chromosome banding patterns in Procyonids, Viverrids and Felids. tein and mitochondrial cytochrome b genes. Zool. Sci. 20, 243– Cytogenet. Cell Genet. 34, 178–192. 264. Xia, X., 2000. DAMBE: Data Analysis in Molecular Biology and Evo- Sato, J.J., Hosoda, T., Wolsan, M., Suzuki, H., 2004. Molecular phylog- lution. Kluwer Academic, Boston. eny of arctoids (Mammalia: Carnivora) with emphasis on phyloge- Yoder, A.D., Burns, M.M., Zehr, S., Delefosse, T., Veron, G., Good- netic and taxonomic positions of the ferret-badgers and skunks. man, S.M., Flynn, J.J., 2003. Single origin of Malagasy carnivore Zool. Sci. 21, 111–118. from an African ancestor. Nature 421, 734–737. Sorenson, M.D., 1999. TreeRot, version 2. Boston Univ., Boston, MA. Yu, L., Li, Q.W., Ryder, O.A., Zhang, Y.P., 2004a. Phylogeny of the SwoVord, D.L., 2001. PAUP*: phylogenetic analysis using parsimony bears (Ursidae) based on nuclear and mitochondrial genes. Mol. (* and other methods). Version 4.0b8. Sinauer Associates, Sunder- Phylogenet. Evol. 32, 480–494. land, MA. Yu, L., Li, Q.W., Ryder, O.A., Zhang, Y.P., 2004b. Phylogenetic rela- Tamura, K., Nei, M., 1993. Estimation of the number of nucleotide tionships within mammalian order Carnivora indicated by substitutions in the control region of mitochondrial DNA in sequences of two nuclear DNA genes. Mol. Phylogenet. Evol. 33, humans and chimpanzees. Mol. Biol. Evol. 10, 512–526. 694–705.